RESEARCHARTICLE The tree species pool of Amazonian wetland forests: Which species can assemble in periodically waterlogged habitats? BrunoGarciaLuize1,2☯*,Jose´LeonardoLimaMagalhães3‡,HelderQueiroz4‡,Maria AparecidaLopes3‡,EduardoMartinsVenticinque5‡,EvlynMa´rciaLeãodeMoraesNovo6‡, ThiagoSannaFreireSilva1,2☯ 1 ProgramadePo´s-Graduac¸ãoemEcologiaeBiodiversidade,InstitutodeBiociências,Universidade EstadualPaulista(UNESP),RioClaro,SãoPaulo,Brazil,2 EcosystemDynamicsObservatory,Departamento a1111111111 deGeografia,InstitutodeGeociênciaseCiênciasExatas,UniversidadeEstadualPaulista(UNESP),Rio a1111111111 Claro,SãoPaulo,Brazil,3 ProgramadePo´s-graduac¸ãoemEcologia,InstitutodeCiênciasBiolo´gicas, a1111111111 UniversidadeFederaldoPara´/EmbrapaAmazoˆniaOriental,Guama´,Bele´m,Para´,Brazil,4 Institutode a1111111111 DesenvolvimentoSustenta´velMamiraua´,Tefe´,Amazonas,Brazil,5 DepartamentodeEcologia,Universidade a1111111111 FederaldoRioGrandedoNorte—UFRN,Natal,RioGrandedoNorte,Brazil,6 DivisãodeSensoriamento Remoto,InstitutoNacionaldePesquisasEspaciais,SãoJose´dosCampos,SãoPaulo,Brazil ☯Theseauthorscontributedequallytothiswork. ‡Theseauthorsalsocontributedequallytothiswork. *[email protected] OPENACCESS Citation:LuizeBG,MagalhãesJLL,QueirozH, LopesMA,VenticinqueEM,LeãodeMoraesNovo Abstract EM,etal.(2018)Thetreespeciespoolof Amazonianwetlandforests:Whichspeciescan WedeterminedthefilteredtreespeciespoolofAmazonianwetlandforests,basedoncon- assembleinperiodicallywaterloggedhabitats? firmedoccurrencerecords,tobetterunderstandhowtreediversityinwetlandenvironments PLoSONE13(5):e0198130.https://doi.org/ 10.1371/journal.pone.0198130 comparestotreediversityintheentireAmazonregion.Thetreespeciespoolwasdeter- minedusingdatafromtwomainsources:1)acompilationofpublishedtreespecieslists Editor:Andre´sViña,MichiganStateUniversity, UNITEDSTATES plusoneunpublishedlistofourown,derivedfromtreeplotinventoriesandfloristicsurveys; 2)queriesonbotanicalcollectionsthatincludeAmazonianflora,curatedbyherbariaand Received:January29,2018 availablethroughtheSpeciesLinkdigitalbiodiversitydatabase.Weappliedtaxonomic Accepted:May14,2018 nameresolutionanddeterminedsample-basedspeciesaccumulationcurvesforbothdata- Published:May29,2018 sets,toestimatesamplingeffortandpredicttheexpectedspeciesrichnessusingChao’s Copyright:©2018Luizeetal.Thisisanopen analyticalestimators.Wereportatotalof3615validtreespeciesoccurringinAmazonian accessarticledistributedunderthetermsofthe wetlandforests.Aftersurveyingalmost70yearsofresearcheffortstoinventorythediversity CreativeCommonsAttributionLicense,which ofAmazonianwetlandtrees,wefoundthat74%theserecordswereregisteredinpublished permitsunrestricteduse,distribution,and reproductioninanymedium,providedtheoriginal specieslists(2688treespecies).Treespeciesrichnessestimatespredictedfromeithersin- authorandsourcearecredited. gledatasetunderestimatedthetotalpooledspeciesrichnessrecordedasoccurringinAma- DataAvailabilityStatement:Allrelevantdataare zonianwetlands,withonly41%ofthespeciessharedbybothdatasets.Thefilteredtree withinthepaperanditsSupportingInformation speciespoolofAmazonianwetlandforestscomprises53%ofthe6727treespeciestaxo- files. nomicallyconfirmedfortheAmazoniantreefloratodate.Thislargeproportionislikelytobe Funding:BGLacknowledgesgrant#2015/24554- theresultofsignificantspeciesinterchangeamongforesthabitatswithintheAmazonregion, 0,SãoPauloResearchFoundation(FAPESP). aswellasinsituspeciationprocessesduetostrongecologicalfiltering.Theprovidedtree JLLMacknowledgesCAPES/PDSE# speciespoolraisesthenumberoftreespeciespreviouslyreportedasoccurringinAmazo- 88881.135761/2016-01andCAPES/Fapespa 1530801.EMVandTSFSacknowledgeCNPq nianwetlandsbyafactorof3.2. productivitygrants#308040/2017-1and#310144/ 2015-9,respectively.Thisworkwassupportedby PLOSONE|https://doi.org/10.1371/journal.pone.0198130 May29,2018 1/13 ThetreespeciespoolofAmazonianwetlands researchgrantsMCT/CNPq/CT-INFRA/GEOMA# Introduction 550373/2010-1and#382728/2010-6andNSF Knowledgeaboutthebiodiversityexpectedforlargerregions,knownastheregionalspecies Dimensions/BiotaFAPESPgrant#2012/50260-6, SãoPauloResearchFoundation(FAPESP).The pool[1],isimportantforinferringevolutionaryprocessesincommunityassembly[2].Empiri- fundershadnoroleinstudydesign,datacollection calstudiesdeterminingthespeciespooloflargeregionsarecentralfordisentanglingthecross- andanalysis,decisiontopublish,orpreparationof scaleprocessesthatshapebiodiversitypatterns[3]butidentifyingthespeciespoolofaregion themanuscript. isnotatrivialtask.Itrequirestheaccumulationofseveralbiodiversitysurveys,well-spaced Competinginterests:Theauthorshavedeclared acrosstheregionandcoveringallpossiblehabitattypes.Theverydefinitionofspeciespoolas thatnocompetinginterestsexist. “thesetofspeciesabletoassemblewithinalocalcommunity”[1,4,5]mustbeconsidered beforeattemptingitsdetermination,asthespeciespoolmaybedefinedintermsofadelimited geographicregion(i.e.unfilteredpool),orregardingaspecifichabitattype(i.e.filteredpool) [1,5]. TheAmazonencompassesmorethanonethirdofallNeotropicalplantdiversity[6,7],dis- tributedamongseveralhabitatswithhighlevelsofheterogeneity[8].Tworecentlypublished checklistsoftheAmazonianflorareportoveralltreespeciesrichnessbetween6727[9]and11 676[10]validspeciesrecordedinherbaria,biodiversityrepositoriesand/orinventories,witha predictedrichnessofc.a.16000treespecies[10,11,12]basedoninventoryobservations.The starkdifferencebetweenchecklistscomesfromamorethoroughtaxonomicreviewperformed by[9],butregardlessofsource,bothlistscanbeconsideredasapproximationsoftheregional unfilteredtreespeciespooloftheAmazonregion,initsbroadestsense[13]. However,theAmazonregioncoversmorethan7millionsquarekilometers,spanning40˚ oflongitude,25˚oflatitude,andanelevationalgradientofc.a.6000m,andmostoftheseveral Amazonianhabitatsremainpoorlysampled[11,14],stronglylimitingourknowledgeofthe trueregionalspeciespool.ItisunreasonabletoexpectthatallAmazoniantreespeciesareable tooccupyeveryenvironment,andthusbepartofthespeciespoolsofallhabitats.Thus,to trulyunderstandtheprocessescontrollingtheassemblyandmaintenanceofAmazondiversity, wemustimproveourknowledgeregardingthefilteredspeciespools[1]ofthediversehabitats comprisingtheAmazonregion. WetlandshavebeenextensivelypresentintheAmazonsinceatleasttheMiocene(30–23 Ma)[15,16],andPleistoceneoceanleveloscillations(2.5Ma)mayhavestronglyinfluenced theirextentanddistributionovertime[17].Wetlandscurrentlycover8.4×105km2ofthe Amazonlowlands(c.a.17%[18]),ofwhichapproximately70%arecoveredbyforests[19]. Totalextentmaybeevenhigher,comprisingupto30%oftheentireAmazonbasin,ifwecon- siderhydromorphicsoilsalongsmallerstreams[20–22].MostAmazonianwetlandsshow monomodalseasonalfluctuationsinwaterstageand/orwatertableheights,knownastheflood pulse[20],whichhasbeeninferredtooccuratleastsincethePaleocene(66Ma)[23]. Hydrologicalseasonalityinfluencesedaphicconditions,leadingtohydrologicalsegregation ofspeciesniches[24,25]asplantsdevelopthephysiologicalandecologicaladaptationsneces- sarytosurviveseveralfloodsanddroughtsduringtheirlifespan[26–31].Thehydrological regimeexperiencedbyeachindividualtreeoccurringintheAmazonianwetlandsdependson localinteractionsbetweenbasinhydrologyandlocalgeomorphology[32],whichcreatestrong gradientsoffloodheightandduration,shapingtreespeciesdiversificationandgeographical distributionacrossscales[22,33–37].Wecanthusconsiderwetlandshabitatsasenvironmental filters,selectingindividualsandspecieswhichcantoleraterecurrentinundationanddrought duringtheirlifespan(e.g.:Hymatanthus [30];Inga[38]),anditisverylikelythatAmazonian wetlandspecieshaveevolvedintoaparticularlyfilteredspeciespool. WhilemosttreediversitystudiesintheAmazonstillfocusonuplandforests,therehasbeen growinginterestinunderstandingtheinfluenceofwater-saturatedenvironmentsonquestions relatedtotreerichness[21,37,39],compositionalpatterns[35,39,40],andphylogenetic PLOSONE|https://doi.org/10.1371/journal.pone.0198130 May29,2018 2/13 ThetreespeciespoolofAmazonianwetlands diversity[38,41,42].AvailabletreespecieslistsforAmazonianwetlandsplacetheeutrophic floodplain(várzea)forestsastherichestwetlandforestsintheworld,with918confirmedtree species[33],andarecentsurveyofBrazilianAmazonianwetlandsraisesthisnumberto1119 treespecies[22],comprising16%ofthe6727treespeciesreportedforoverallAmazonlow- landforests[9].Furthermore,basedon542taxa(speciesandmorpho-species),threemainbio- geographicregionsaresupportedbytreespeciescompositionalchangesalongtheBrazilian Amazonrivermainstem[35].Itisthusclearthatweneedamorecomprehensiveknowledge ofthefilteredspeciespoolabletocolonizethesehabitats,tobetterunderstandthehydrological dimensionofnichesoccupiedbyAmazoniantreespecies[24,25]anditsroleintheassembly andevolutionofAmazonrainforests. Here,weprovidethemostcomprehensiveestimatetodateofthefilteredtreespeciespool abletoassembleinAmazonianwetlands,combiningtreespeciesrecordsfromherbariadata- basesandpublishedandunpublishedtreespeciessurveysfromdifferenttypesofAmazonian wetlandforests.WealsodiscussthepossibleroleofwetlandsinmaintainingAmazontree diversity,andofferapredictiontotheexpectednumberofspeciescomprisingthetotalfiltered treespeciespoolthatcansurviveinwetlandenvironments,assessinghowitcomparestothe knownAmazontreefloraandpredictedbasinwidediversity.Finally,wediscusscurrentlimi- tationsandbestpracticesforincreasingourbiogeographicalknowledgeofthemosttreespe- ciesrichanddiversewetlandforestsintheworld. Materialsandmethods Datasets Ourfirstdatasetcomprisesareviewofpublishedtreespecieslists(TSL)fromtreeplotinvento- riesand/orfloristicsurveysconductedinAmazonianwetlandforests(Fig1),complemented byonepreviouslyunpublishedprimaryinventoryofourown(S1Table).ToconstructTSL, weonlyconsideredstudiesthatreportedcompletespecieslists,foranyAmazonianwetland type[20]. Ourseconddatasetwasbuiltbyqueryingbotanicalcollections(BC)madeinAmazonian wetlandforests,curatedbyherbaria(Fig1,S2Table)andincludedintheSpeciesLinkdigital biodiversitydatabase(http://www.splink.org.br).Wequerieddigitizedvoucherlabelsusing thefollowingkeywords:“Alagada”; “Alagado”; “Alagável”; “Aluvial”; “Alluvial”;“Área Úmida”; “Brejo”; “Chavascal”;“Flooded”;“Flood”;“Floodplain”; “Hidromórfico”; “Hydromorphic”; “Igapó”; “Inundada”; “Inundável”; “Restinga”; “Tahuampa”; “Várzea”. Wethenmergedall botanicalrecordsreturnedforeachkeywordandfilteredtheserecordstoincludeonlyAngio- spermspeciesandonlyspecimenscollectedintheAmazoniasensu-latissimoregion,asdefined by[43](Fig1). Taxonomicstandardization Validcanonicalnamesforspecieswereachievedbyperformingtaxonomicnameresolution forbothspeciesdatasets,usingtheTaxonomicNameResolutionService—TNRSV.4.0online platform[46].WesetTNRStoperformnameresolutionwithoutallowingpartialmatches, andwithaminimummatchthreshold>0.85.Theauthoritysourcesconsultedwere,inorder ofrelevance,TROPICOS(http://www.tropicos.org)andTHEPLANTLIST(http://theplantlist.org), lastupdatedonAugust2015(fordetailssee:http://tnrs.iplantcollaborative.org).FortheTSL dataset,afterperformingtaxonomicnameresolution,wefilteredtheresultingrecordsto removefamiliesknowntocompriseonlynon-treelifeforms,andweassumedallremaining recordsafterfilteringcorrespondedtotreespecies.ThefilteredrecordsfromBCdatasetwere matchedtothemostrecentAmazontreeflorachecklist[9],retainingonlyspeciesnames PLOSONE|https://doi.org/10.1371/journal.pone.0198130 May29,2018 3/13 ThetreespeciespoolofAmazonianwetlands Fig1.LocationofpublishedspecieslistsandherbariarecordsreportingtreespeciesonAmazonianwetlandsforests.Thereddotsarethelocationoftree specieslists(TSL)frombotanicalinventoriesonAmazonianwetlands,bluedotsarethevoucherspecimensfrombotanicalcollections(BC).TheAmazonia sensu-latissimoregionisdefinedin[43],wetlandareaswereobtainedfrom[44],andtheclassificationofmajorAmazonianrivertypesisgivenby[45]. https://doi.org/10.1371/journal.pone.0198130.g001 confirmedbytaxonomicspecialistsasvalidspeciesnamesandhavingatreelifeform(i.e.lig- neoustrunkreaching10cmDBH). Richnessestimation WeusedtheTSLandBCdatasetstobuildseparatespecies-by-sampling-unitincidencematri- ces,aggregatingincidencebystudyforTSL,andbyyearofcollectionforBC.Weusedthe resultingmatricestoassessthechronologicalorderofincidenceofeachrecordedspecies, buildingacumulativespeciescollector’scurveusingthe‘vegan’package[47]andtoobtaining therespectivesample-basedspeciesaccumulationcurvesforeachdataset[48].Wethenused thesample-basedcurvestopredicttheexpectedspeciesrichnessifcollectioneffortsweredou- bled.Theinferredandestimatedsample-basedaccumulationcurvesandpredictionsofspecies richnesswerecalculatedusingrarefactionandextrapolationfunctionsforincidencedatapro- videdby[48],usingthe‘iNEXT’package[49].AllanalyseswereperformedinR3.3.2.[50]. PLOSONE|https://doi.org/10.1371/journal.pone.0198130 May29,2018 4/13 ThetreespeciespoolofAmazonianwetlands Resultsanddiscussion DeterminingthefilteredspeciespoolofAmazonianwetlands Intotal,wereviewed69studiesreportingtreespecieslistsforinventoriesconducedonAmazo- nianwetlandforests(S1Table),ofwhich16(~20%)didnotincludeacompletelistofspeciesand couldnotbeaddedtotheTSLdataset.Fromthe53studiesincludedinTSL,werecovered21446 recordscomprising2688validtreespeciesnames(S3Table).Fromthese,weestimatethat3380 (lower95%=3305,upper95%=3455)treespecieswouldberecordedforAmazonwetlandfor- estsifsamplingeffortwasdoubled(Fig2A).Neitherthecollector’scurve,northeestimatedsam- ple-basedspeciesaccumulationcurveshowedsignsofreachinganasymptote(Fig2A),evenafter almost70yearsofinventoriesbeingconductedinAmazonianwetlandforests. Weretrieved231119plantoccurrencerecordsfromtheSpeciesLinkdatabase.Afterfiltering forAngiospermsintheAmazonregion,performingtaxonomicnameresolutionandmatching againstthereferencetreespecieslists,weretained20902recordsfor2408validtreespecies names(BCdataset—S2TableandS3Table),lowerthantheobservedorexpectednumberof treespeciesobtainedfromtheTSLdataset.FortheBCdataset,wepredictedanexpectedrich- nessof2938treespecies(lower95%=2867,upper95%=3009)toberecordedforAmazo- nianwetlandforestsifcollectioneffortsweredoubled(Fig2B). PoolingtogethertheTSLandBCdatasetsconfirmedatotalof3615validtreespecies,com- prising42348recordsoftreesoccurringinAmazonianwetlandforests(S3Table),ahigher richnessthantheexpecteddoubling-effortpredictionsfromeitherisolateddataset.Thetwo datasetsshared1481(c.a.41%)treespecies,with1207(c.a.33%)onlyrecordedbyTSLand 927(c.a.26%)treespeciesonlyrecordedbyBC. Scopeandlimitationsofthedeterminedtreespeciespool ThedeterminedtreespeciespoolofAmazonianwetlandforestscomprises3615validspecies, encompassingenvironmentalconditionsfoundbetweendiversewetlandtypes[20].Thisisthe Fig2.Cumulativecollector’scurveandsample-basedspeciesaccumulationcurvefortreespeciesinAmazonianwetlands.(A)Treespecieslists(TSL)ordered from1950to2017(seeS1Tableforalistofreviewedstudies).(B)Botanicalcollections(BC)from1857to2016(seeS2Tableforalistofherbariawhererecordsare available).Thedotsrepresentthecumulativenumberofspecies,thesolidredlineistheresultofrandominterpolationofthesepoints,andthedashedredlineisthe predictednumberofrecordedspecieswithincreasedeffort[48].Thegrayareadenotesthe95%confidenceintervaloftheestimatedcurves. https://doi.org/10.1371/journal.pone.0198130.g002 PLOSONE|https://doi.org/10.1371/journal.pone.0198130 May29,2018 5/13 ThetreespeciespoolofAmazonianwetlands mostcomprehensiveestimatetodateoftheAmazoniantreespeciespoolthatcansurvive underextremehydrologicalconditions.AlthoughthesamplingeffortdevotedtoAmazonian uplandforestsiscurrentlyfourtimeshigherthantowetlandforests[11,23],ourtreespecies listrepresents53%ofallthe6727treespeciesconfirmedfortheentireAmazonregion[9]. Assumingthistobeanaccurateestimateofthetrueproportion,Amazonianwetlandscould harborc.a.8500ofthe16000treespeciesexpectedtocomprisethetotalAmazoniantreeflora [11]. Mostlikely,othertreespeciesreportedfortheAmazonmayalsooccurinhydromorphic environments,buthavenotyetbeenrecordedinAmazonianwetlands.Forinstance,theaver- agecollectiondensityrecoveredbyus(TSL+BC)is0.020recordsper100km2ofAmazonian wetlands,whenconsideringthe2.1millionkm2estimateof[20],or0.050recordsper100km2 ifconsideringthemorerestrictive840000km2mappedby[18].Thesesamplingdensitiesare threeordersofmagnitudelowerthantheobserveddensityof10recordsper100km2forAma- zonianforestsingeneral[9,51].Forthisreason,wealsoexpectthatanimportantportionof treespeciesoccurringinAmazonianwetlandsmaynotbeyetknowntoscience.Forexample, fromthe173treespeciesdiscoveredintheAmazonduringthefirstdecadeofthe21stcentury [52],only21(12%)wereidentifiedinourestimatedspeciespool,andofthese,onlysixholo- typespecimensseemtocomefromvoucherscollectedinAmazonianwetlandhabitats.We thusemphasizethedireneedformoreintensiveandcomprehensivesamplingoftheAmazo- nianwetlandenvironments. Asecondlimitationofthepresentlistisintroducedbythebiastowardsspecificwetland typeswithintheAmazon.BiodiversityassessmentsintheAmazonandelsewherearegenerally biasedtowardsmajorurbancentersandalongmajorriversorroadways[9,53],andthisbiasis showntowardsinventoriesofcertaintypesoffloodplainforests.Thecoverageofwetlandhabi- tattypesandspeciesoccurrencesrecordedinourTSLandBCdatasetsshow,aspreviouslyrec- ognizedby[23],thateutrophicfloodplainforests(várzeas)alonglarge“white-water”riversare themostsampledwetlandforesttypeacrosstheAmazon.MostoftheAmazonianhumanpop- ulationandmajorurbancentersareadjacenttotheseareas,andwefoundthelargestdensities ofbotanicalrecordsalongtheAmazonasandNegrorivermainstems,nearmajorurbancen- terswithwell-establishedresearchinstitutions(e.g.:Bele´m,Manaus,Tefe´,Iquitos).Amuch lowerrecorddensitywasobservedalongthefloodplainwetlandsofothermajorAmazon tributaries(e.g.:Putumayo-Ic¸a´;Jurua´;PurusandMadeira),orinriparianforestsalonginterflu- vialareasoftheAmazonlowlands. Athirdlimitationisthatwecouldnotuseoneineveryfour(25%)publishedtreesurveys conductedinAmazonianwetlandforests,astheauthorsdidnotincludeexplicitandcomplete specieslistsinthepublications.Althoughthe21stcenturyhasseentheriseofcollaborativenet- works,andcomprehensivechecklistsforNeotropicalforestsprovidelargeamountsofvaluable information,westillneedadeeperculturalshiftamongresearchers,favoringdatasharingand transparency,ifwearetoimproveourcombinedknowledgeoftropicaltreebiodiversity[54]. Itissurprisingthatthetwodatasetsweinvestigatedsharedlessthanhalfofthetotalnumberof validtreespeciesrecorded,aswewouldexpectcompleteoverlapunderanidealscenario whereatleastonevoucherspecimenwasdepositedforeachspeciesrecordedineachreviewed inventory(withvouchersproperlydigitizedandmadeavailableonlineinherbariadatabases). However,althoughmostpublishedinventorystudiesclaimedtohavedepositedvoucherspeci- mensfortheirsampledplots,wewereunabletofindnearlyathirdofthespeciesreportedfor inventoryplotsinthedigitizedherbariasources.Veryoften,easilyrecognizablespeciesand specimenswithoutfertilestructuresarenotincludedinvouchercollections,creatinga“data void”intheherbariarecords[51].Thus,inpractice,inventoriesandisolatedbotanicalcollec- tionsprovidecomplementaryfloristicinformationforassessingtreespeciesdiversity.This PLOSONE|https://doi.org/10.1371/journal.pone.0198130 May29,2018 6/13 ThetreespeciespoolofAmazonianwetlands reinforcestheneedforincludingthecompletespecieslistsinpublishedinventoriesandshows thatscientistsneedtokeepperformingbothtypesofstudiesifwearetoincreaseourknowl- edgeoftheAmazonwetlandtreediversity. Finally,amorecomprehensiveknowledgeoftheAmazonwetlandstreespeciespoolcanbe achievedthrougheffortsinreducingotherbiologicalshortfalls(sensu[55]).Forinstance,the uncertaintyregardingactuallife-form(i.e.:tree)oftherecordedplantspecies(“Raunkiaeran shortfall”),andthelackofvoucherdeterminationsandtaxonomicreviewsformostherbaria records(“Linneanshortfall”),resultedintheremovalofc.a.25000recordsand6000species namesoriginallypresentintheBCdatasetaftertaxonomicstandardizationandmatchingto thetreespecieslistof[9].Furthermore,manysamplesdidnotincludeinformationonhabitat conditions,precludingadetailedassessmentofspeciesoccurrencebywetlandtype(e.g.:vár- zea,igapó,campinas,tidalvárzeas).Moreeffortsshouldbemadetoensureforthcomingbotan- icalcollectionsandinventoriesexplicitlyincludelifeformandspecifichabitatconditions,as wellasotherecologicallyrelevantinformation. HowdoestheAmazonianwetlandspeciespoolcomparetothebasin-wide speciespool? ThetreespeciespoolofAmazonwetlandscomprised104botanicalfamiliesdistributedinto 689genera,withelevenfamilieshavingmorethan100treespecieseach.Leguminosae(578 treespecies),Rubiaceae(220treespecies),Annonaceae(182treespecies),Lauraceae(175tree species),andMyrtaceae(155treespecies)werethemostdiversetreefamiliesinAmazonian wetlandforests,comprisingtogether36%oftheAmazonianwetlandstreespeciespool.The tenrichestfamiliesinAmazonianwetlandsaccountedfor53%oftheentirespeciespool (Table1). Although69treefamilieshadhalformoreoftheirAmazoniantaxaoccurringinAmazon wetlands,includingsomeoftherichestwetlandfamilies(Leguminosae,Euphorbiaceaeand Moraceae,Table1),wedidnotfindanywetlandrecordsfor15familieswithknownoccur- renceinAmazonforests.Overall,c.a.51%oftheAmazoniantreespecieswithineachfamily occurredinwetlandhabitats,buttherewerenoticeabledifferencesinrankorderandpercent- ageofsharedspeciesbetweenthetenrichestwetland-occurringfamiliesandtheirrespective richnessrankingwithintheoverallAmazonflora,asgivenby[9](Table1). Table1. TreespeciesrichnessforthetenrichestbotanicalfamiliesfoundinAmazonianwetlandscomparedwiththeirrichnessrankingaccordingtotheAmazon treeflora. Family Richnessrankingfor 1Richnessrankingfor Numberofvalidtree Numberofvalidtree Percentofspecies Amazonianwetlandstree Amazoniantreeflora speciesinAmazon speciesinentireAmazon occurringinwetlands speciespool Wetlands flora1 (%) Leguminosae 1 1 578 1042 55 Rubiaceae 2 5 220 338 65 Annonaceae 3 4 182 388 46 Lauraceae 4 2 175 400 43 Myrtaceae 5 3 155 393 39 Melastomataceae 6 6 136 263 51 Chrysobalanaceae 7 7 132 256 51 Sapotaceae 8 8 128 244 52 Euphorbiaceae 9 11 114 160 71 Moraceae 10 13 112 147 76 1Following[9]. https://doi.org/10.1371/journal.pone.0198130.t001 PLOSONE|https://doi.org/10.1371/journal.pone.0198130 May29,2018 7/13 ThetreespeciespoolofAmazonianwetlands Fig3.Per-genusproportionofAmazoniantreespeciesoccurringandnotoccurringinwetlands.Proportionsarecalculatedforthe803generalistedtheAmazon treespecieschecklist[9]andrankedfromhighertolowerproportionofspeciesonwetlands. https://doi.org/10.1371/journal.pone.0198130.g003 Atthegenuslevel,221generaintheAmazontreechecklist[9]hadallitsknownspecies recordedintheAmazonwetlandstreespeciespool(Fig3).However,manyofthesegenera (124)hadonlyasingleacceptedspeciesoccurringintheAmazon,withonlyeightgenerahav- ing10ormoreknownspecies(max.26species).Conversely,201generalistedontheAmazon treechecklist[9]hadnospeciesrecordedinAmazonianwetlands(Fig3).Therichestgenusin AmazonwetlandsisInga(85treespecies),followedbyLicania(69species),Miconia(69spe- cies),Pouteria(69species),andEugenia(59species). TheecologicalandevolutionaryroleofAmazonianwetlands ThefilteredtreespeciespoolforAmazonianwetlandforestsincludesalmostallbotanicalfami- liesknowntooccurinAmazonforests.Itiscomparabletothe3389treespeciesacknowledged fortheentireBrazilianAtlanticForest[56],oneofthemostbiodiverseNeotropicalbiomes. Onepossibleexplanationforthisrichnessisthat,asAmazonianuplandandwetlandareasare contiguoushabitatsknowntohaveaninterchangeableflora[34,40],wecanexpectahigh degreeoflateralmigrationsamongthesehabitats,withalargeproportionoftreespeciesin eachlineagereachingandeventuallyadaptingtobothfloodedandnon-floodedforestedhabi- tats.Still,differentpatternsmightalsobeplausible.Forinstance,thecontributionoftreespe- ciesoccurringinAmazonianwetlandstothetotaldiversityoftheAmazon-centeredgenus suggestsometaxahaveevolvedahighdegreeofinsituspecializationonwetlands,onlythen colonizinguplandhabitats.Despitethehighlikelihoodthatatreespecieswillreachwetland habitatswhenmigratingacrosstheAmazonlandscape,manyAmazoniantreespeciesdonot PLOSONE|https://doi.org/10.1371/journal.pone.0198130 May29,2018 8/13 ThetreespeciespoolofAmazonianwetlands showpreferenceforfloodedhabitats;c.a.64%ofthe4963treespeciesrecordedinATDN database[11],withonly68ofthe600mostcommontreespeciesoccurringinwhite-water Amazonfloodplainforestseemingtobehabitatendemics[34].Assessingphylogenetichistory andtherelativecontributionofeachdirectionofmigrationtodiversificationcouldgiveus importantinsightontheoriginandevolutionaryhistoryofseveralimportanttaxaintheAma- zontreeflora,andtheroleofstrongenvironmentalfilteringandhydrologicalnichespecializa- tioninthisprocess,ashasbeenshownforBrazilianCerradospeciesinrelationtofire disturbance[57]. Growingevidencesuggeststhatitisreasonabletothinkofatreespeciespoolcomprisedby theentireAmazonregion[13],buttheroleofecologicalfilteringintheassemblyoflocalcom- munitiescannotbeexcluded[58].ThecontinentaldimensionsoftheAmazonbiomeandthe virtuallackofgeographicbarriersforplantspeciesacrossthelowlandsimpliesfewdispersal limitationsfortreespecies[13].Newenvironmentalconditionsarereachedwhenspecies expandtheirdistributions,andthisfloristicinterchangebetweenwetlandanduplandhabitats mightmodulatesource-sinkpopulationdynamicsacrossmarginalhabitats.Atecologicaltime- scales,source-sinkdynamicswillaffectpopulationregulationandspeciescoexistence[59,60]; overevolutionarytimescales,itwillselectecotypesmorepronetocolonizecertainhabitats, leadingtogeneticandmorphologicaldifferentiationamongpopulations[30,58,61].Inthis context,althoughtheAmazonianhydrologicalgradientsaremoreidiosyncraticthanthecon- spicuousandwidelydiscussedtemperaturegradientsalongAndeanmountainslopes,thereis ampleevidenceforselectivepressuresactingonthehydrologicalnichedimensionofAmazo- niantreespecies,stronglyaffectingvegetationdevelopmentandthedistributionofspecies diversityacrosstheregion[11,39,62].Therefore,theselowlandhydrologicalgradientsare verylikelytohavehadastronghistoricalroleontreespeciesdiversification,rangeexpansion [34,38,42,63],andlocalcommunityassembly[37,39]. Conclusions WeshowthatthetreespeciespoolofAmazonianwetlandscomprises53%(3615)ofthecon- firmedtreespeciesoccurringintheoverallAmazon,raisingpreviousrichnessestimatesbya factorof3.2.Itisverylikelythatmanyofthesespecieswillalsooccurinotherforestedhabitats, orevenotherNeotropicalregions.AlargeportionoftheNeotropicalplantdiversityisencom- passedbyAmazon-centeredtaxaandunderstandingtheirevolutionaryandecologicalhisto- riescanimproveourknowledgeofthedevelopmentofthishyperdiversebiogeographicrealm. GeographicalbarriersforplantdispersalaremostlyabsentintheAmazonregion,whichis insteadcharacterizedbyamosaicofhabitattypesandenvironmentalgradients,includingwet- landhabitatsthathavebeenpervasivelypresentsincebeforetheAndeanuplift.Furtherstudies thatcandisassembleandthencontrasttheAmazontreefloraintothefilteredspeciespools associatedwitheachhabitattypearenecessarytoopennewavenuesforexploringtheecologi- calandgeographicdistributionofAmazoniantreespecies,functionaltypes,andlineages,and unveiltherelativeroleofdispersalandenvironmentalfilteringoncommunityassemblyand ontheoriginsandmaintenanceofspeciesdiversityovertime. Supportinginformation S1Table.Referencelistforreviewedtreespecieslists. (XLSX) S2Table.ListofherbariaavailableonSpeciesLinkthatcontributedwithrecords. (XLSX) PLOSONE|https://doi.org/10.1371/journal.pone.0198130 May29,2018 9/13 ThetreespeciespoolofAmazonianwetlands S3Table.ChecklistoftheAmazonianwetlandstreespeciespool. (XLSX) Acknowledgments WethankLuciaG.LohmannandAnaCarolinaC.Carnavalforhelpfuldiscussionswhenplan- ningthescopeofthismanuscript. AuthorContributions Conceptualization:BrunoGarciaLuize,EduardoMartinsVenticinque,EvlynMa´rciaLeãode MoraesNovo,ThiagoSannaFreireSilva. Datacuration:BrunoGarciaLuize. Formalanalysis:BrunoGarciaLuize,ThiagoSannaFreireSilva. Fundingacquisition:HelderQueiroz,MariaAparecidaLopes,EduardoMartinsVenticinque, EvlynMa´rciaLeãodeMoraesNovo,ThiagoSannaFreireSilva. Investigation:BrunoGarciaLuize,Jose´LeonardoLimaMagalhães. Methodology:BrunoGarciaLuize. Projectadministration:HelderQueiroz,MariaAparecidaLopes,EduardoMartinsVenticin- que,EvlynMa´rciaLeãodeMoraesNovo,ThiagoSannaFreireSilva. Supervision:ThiagoSannaFreireSilva. Validation:BrunoGarciaLuize. Visualization:BrunoGarciaLuize,EduardoMartinsVenticinque,ThiagoSannaFreireSilva. Writing–originaldraft:BrunoGarciaLuize,ThiagoSannaFreireSilva. Writing–review&editing:BrunoGarciaLuize,Jose´LeonardoLimaMagalhães,Helder Queiroz,MariaAparecidaLopes,EduardoMartinsVenticinque,EvlynMa´rciaLeãode MoraesNovo,ThiagoSannaFreireSilva. References 1. CornellHV,HarrisonSP.WhatAreSpeciesPoolsandWhenAreTheyImportant?AnnuRevEcolEvol Syst.2014;45:45–67.https://doi.org/10.1146/annurev-ecolsys-120213-091759 2. CarstensenDW,LessardJP,HoltBG,KrabbeBorregaardM,RahbekC.Introducingthebiogeographic speciespool.Ecography(Cop).2013;36:1310–1318.https://doi.org/10.1111/j.1600-0587.2013. 00329.x 3. RicklefsRE,HeF.Regioneffectsinfluencelocaltreespeciesdiversity.ProcNatlAcadSci.2016;113: 674–679.https://doi.org/10.1073/pnas.1523683113PMID:26733680 4. SrivastavaDS.Usinglocal-regionalrichnessplotstotestforspeciessaturation:pitfallsandpotentials.J AnimEcol.1999;68:1–16.https://doi.org/10.1046/j.1365-2656.1999.00266.x 5. ZobelM.Thespeciespoolconceptasaframeworkforstudyingpatternsofplantdiversity.JVegSci. 2016;27:8–18.https://doi.org/10.1111/jvs.12333 6. GentryAH.NeotropicalFloristicDiversity:PhytogeographicalConnectionsBetweenCentralandSouth America,PleistoceneClimaticFluctuations,oranAccidentoftheAndeanOrogeny?AnnMissouriBot Gard.1982;69:557.https://doi.org/10.2307/2399084 7. AntonelliA,Sanmart´ınI.WhyaretheresomanyplantspeciesintheNeotropics?Taxon.2011;60: 403–414.https://doi.org/10.2307/41317138 8. OlsonDM,DinertsteinE,WikramanayakeED,BurguessND,PowellGVN,UnderwoodEC,etal.Ter- restrialecoregionsoftheworld:anewmapoflifeonEarth.Bioscience.2001;51:933–938. PLOSONE|https://doi.org/10.1371/journal.pone.0198130 May29,2018 10/13
Description: